Title

How Many Runs? Analytical Method for Optimal Scenario Sampling to Estimate the Variance of Travel Time Distributions in Vehicular Traffic Networks -- Winner of the 2014 TRB Ryuichi Kitamura Best Paper Award

Time

1:40 p.m. – 3:00 p.m.

Location

1605 Tilia, Room 1103, West Village

Abstract

Scenario-based approaches provide an effective and practical method for capturing the probabilistic nature of travel time in a traffic network. Scenarios that represent daily roadway conditions are generated by identifying various demand- and supply-side factors that affect travel time variability and sampling a set of mutually consistent combinations of the associated events. The sampled scenarios are then evaluated using traffic simulation models to obtain travel time distributions, which provide a basis for extracting a wide range of reliability performance metrics. A key question under this framework pertains to the number of input scenarios needed to achieve the best estimators of the reliability measures of interest given a limited computational budget. Given a stratification of the entire domain of daily scenarios into distinct scenario categories (or strata), the study addresses the optimal sample size allocation problem in connection with stratified sampling. Existing sample allocation schemes, e.g., Neyman’s, are optimized for estimation of the mean. However, dispersion measures such as variance and standard deviation are of greater interest in travel time reliability studies. Thus, this study explicitly specifies the optimal allocation scheme for the estimation of the variance of travel time. A specific characteristic observed in travel time data, namely, a strong positive correlation between standard deviation and mean, is used to develop an analytical formula that approximates the sample’s variance and to derive an approximate solution for optimal allocation for estimating variance. The proposed method is validated using a simulation study and compared with other allocation methods in terms of the estimation of various reliability measures.

(Joint work with Professor Hani S. Mahmassani, Northwestern University)

Biographical Sketch

Dr. Jiwon Kim is an assistant professor in Transport Group in the School of Civil Engineering, the University of Queensland, which she joined in June 2014. She received her Ph.D. in Transportation Systems Analysis and Planning from Northwestern University in 2014 for her thesis “Travel Time Reliability of Traffic Networks: Characterization, Modeling, and Scenario-based Simulation.”

Dr. Kim’s research is broadly in the area of modeling and analysis of urban transport systems, with an emphasis on travel time reliability analysis, traffic flow theory, large-scale dynamic network modeling and traffic simulation, and decision support systems for real-time traffic management and operations. Her current research focuses on the application of probabilistic modeling and machine learning methods to study relationships among events, traffic, and control actions in urban networks to help address complex traffic control and management issues in the dynamic ITS environment.

Dr. Kim was awarded the Inaugural Greenshields Prize by the Committee on Traffic Flow Theory and Characteristics (AHB45), Transportation Research Board (TRB), in 2011 and received the Best Paper Award from the TRB Transportation Network Modeling Committee (ADB30) in 2014.

http://www.civil.uq.edu.au/jiwonkim